The fit based defect inspection is an evaluation that is supposed to find local deviations of a measurement in comparison to a reference. It can be used to find defects on a sample such as cracks, edged chips, digs, scratches, etc. Therefore, a reference form (the ideal sample form without any defects) has to be known and provided as a mathematical function. Parameters of the mathematical function can be left open to be adapted by a fit algorithm. Thus, although keeping the basic shape, the reference form can be adapted to the measured sample. Deviations between the adapted reference form and measurement are then determined, clustered and classified.
The basic steps of the entire procedure are:
1. Determine the area of the sample within the measurement and cut out measurement data apart from the sample
2. Adapt the parameters of the function that represents the reference form so that the reference form fits best the shape of the measured sample
3. Determine deviations between fit and measurement
4. Cluster deviations, assess and classify them
The user can set the threshold at which a deviating measurement point is considered as a defect point. Deviations smaller than this threshold are not considered as defects. One might be only interested in defects whose lateral expansion is greater than a set threshold. This property can be used in order to consider only deviations (areas of deviation) whose maximum lateral expansion is greater than the set value.
Found defects are classified in terms of their shape, particularly in terms of the aspect ratio between their longest lateral expansion and their expansion that is orthogonal to the longest expansion. Two classes are available: Digs and scratches. Defects with a bigger aspect ratio than individually defined by the user are classified as scratches, otherwise as digs.
We certainly have a solution for your specific task. Do not hesitate to contact us if you have any questions. Our experts will be glad to take care of your needs and work out individual solutions for you.
Interested in latest posts of this series? Click below.
Part 2/7: Sample Classification function offers a sophisticated way of classifying measured samples
Part 3/7: Dual Scan Mode for high-speed measurement of features with a large discontinuous step
Part 5/7: Individual Execution of Tasks and Evaluations for Recipe Adjustment